Complexation of Independent Biometric Features in People Recognition with Quadratic Forms, Perceptrons and Functional Hee-module
Abstract
Introduction: Static biometric features are not a secret and can be falsified, so the search for effective methods of authenticating people by their dynamic biometric characteristics is a very important problem. Purpose: The aim is to develop more reliable methods for one- and multi-factor biometric authentication using uninformative features. Results: A series of numerical experiments were conducted on the basis of biometric data of a signature, keyboard handwriting, faces and voices of people using perceptrons networks, quadratic forms and functional Hee-module. An error level for the user verification by handwriting dynamics is about 1%, an error for the user verification by keyboard signature and handwriting dynamics is about 0.31%, face image gives an error level less than 0.5%, face and keyboard signature gives an error level less than 0.1%, using 3- and 4-factor verification gives 0,54-0,01% error level. Practical relevance: Methods of two- (without voice features), three- and four-factor user verification discussed in the paper can be used in practice for the implementation of a remote authentication function. Forging features of more than two kinds of images can be considered practically impossible.Published
2017-02-20
How to Cite
Sulavko, A., Eremenko, A., Tolkacheva, E., & Borisov, R. (2017). Complexation of Independent Biometric Features in People Recognition with Quadratic Forms, Perceptrons and Functional Hee-module. Information and Control Systems, (1), 50-62. https://doi.org/10.15217/issn1684-8853.2017.1.50
Issue
Section
Information coding and transmission